The subjective memory complaints reported by older adults involve memory for everyday activities, such as forgetting where they placed items, forgetting directions to familiar locations, and repeating stories. Yet, age-related changes in memory in the context of naturalistic everyday events have gone surprisingly unstudied. Recently, we have begun to understand how people build and update representations of events they experience or read about. Activity is dynamic, so the representation of what is happening now (i.e., situation model) must be updated constantly. Decreased memory updating and capacity are associated with aging; thus, older adults may update situation models differently than do young adults. Situation models are updated when a change has occurred along at least one of several dimensions, including characters, goals, space, and time. One possibility is that all situation model updating is incremental in which only information relevant to the changing dimension is updated. Alternatively, situation models may be globally updated. In this view, dynamic activity is segmented into discrete events that are represented in the situation model. Activity is segmented at an event boundary, and when a new event begins, the entire model is updated-not just the information that changed. Given that little evidence exists for global updating, this project aims to test whether (1) situation models are updated globally and (2) the updating process is affected by age. To assess these aims, young and older adults will read narrative texts. Memory will be probed following an event boundary triggered by a change along one dimension (e.g., space). If we find evidence of global updating, then all information related to the previous event will be less available after an event boundary. However, if updating is incremental, then we should find that only information associated with the changed dimension is less available. Finally, while participants read narratives, functional magnetic imaging will be used to look at age-related differences in phasic activity in response to memory probes. If situation models are updated globally at event boundaries, then all probes should selectively activate the medial temporal lobes-regions associated with retrieval from long-term memory. If situation models are updated incrementally, then the changed-dimension probes should activate the medial temporal lobes, whereas the unchanged-dimension probes should activate prefrontal cortex-region associated with retrieval from a situation model. This project has implications for theories of comprehension as well as cognitive aging interventions.